Rational engineering of Saccharomyces cerevisiae towards improved tolerance to multiple inhibitors in lignocellulose fermentations

RESEARCH

Rational engineering of Saccharomyces cerevisiae towards improved tolerance to multiple inhibitors in lignocellulose fermentations

Bianca A. Brandt¹, Maria D. P. García‑Aparicio², Johann F. Görgens² and Willem H. van Zyl^1*

Abstract

Background: The fermentation of lignocellulose hydrolysates to ethanol requires robust xylose‑capable Saccharo- myces cerevisiae strains able to operate in the presence of microbial inhibitory stresses. This study aimed at developing industrial S. cerevisiae strains with enhanced tolerance towards pretreatment‑derived microbial inhibitors, by identify‑

ing novel gene combinations that confer resistance to multiple inhibitors (thus cumulative inhibitor resistance pheno‑

type) with minimum impact on the xylose fermentation ability. The strategy consisted of multiple sequential delta‑

integrations of double‑gene cassettes containing one gene conferring broad inhibitor tolerance (ARI1, PAD1 or TAL1) coupled with an inhibitor‑specific gene (ADH6, FDH1 or ICT1). The performances of the transformants were compared with the parental strain in terms of biomass growth, ethanol yields and productivity, as well as detoxification capaci‑

ties in a synthetic inhibitor cocktail, sugarcane bagasse hydrolysate as well as hardwood spent sulphite liquor.

Results: The first and second round of delta‑integrated transformants exhibited a trade‑off between biomass and ethanol yield. Transformants showed increased inhibitor resistance phenotypes relative to parental controls specifi‑

cally in fermentations with concentrated spent sulphite liquors at 40% and 80% v/v concentrations in 2% SC media.

Unexpectedly, the xylose fermentation capacity of the transformants was reduced compared to the parental control, but certain combinations of genes had a minor impact (e.g. TAL1 + FDH1). The TAL1 + ICT1 combination negatively impacted on both biomass growth and ethanol yield, which could be linked to the ICT1 protein increasing transfor‑

mant susceptibility to weak acids and temperature due to cell membrane changes.

Conclusions: The integration of the selected genes was proven to increase tolerance to pretreatment inhibitors in synthetic or industrial hydrolysates, but they were limited to the fermentation of glucose. However, some gene com‑

bination sequences had a reduced impact on xylose conversion.

Keywords: Lignocellulose, Saccharomyces cerevisiae, Microbial inhibitors, Spent sulphite liquor

© The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Background

The increased pressure towards decreased carbon emissions has spurred the development of lignocellu- lose-derived biofuels production as replacement for conventional fossil fuels [1, 2]. Fermentation serves as bioconversion to alcohols of sugar hydrolysates derived from polysaccharide-rich lignocellulose biomass. How- ever, a major challenge linked to hydrolysis-fermentation

Open Access

*Correspondence: whvz@sun.ac.za

1 Department of Microbiology, Stellenbosch University, Private Bag X1, Stellenbosch 7602, South Africa

Full list of author information is available at the end of the article

of lignocellulose biomass is the recalcitrant nature of the material to enzymatic conversion [3, 4]. Physicochemi- cal pre-treatment is thus required to disrupt the compact crystalline structure and allow enzymatic access to the polysaccharides within, to release fermentable sugars [3, 5, 6]. The majority of such pre-treatment methods result in significant quantities of degradation products being formed, which have inhibitory effects of subsequent bio- logical conversions [5, 7–11].

Saccharomyces cerevisiae cannot naturally utilize xylose, the most abundant pentose sugar within lignocel- lulosic hydrolysates [12–14]. Although co-fermentation of glucose and xylose remains a challenge, advances in strain development have resulted in the development of industrial S. cerevisiae strains with xylose-utilizing capacity. Through metabolic engineering, heterologous xylose catabolic pathways such as the fungal oxidoreduc- tive pathway (XR-XDH) [13, 15, 16] or a bacterial xylose isomerase (XI) [17, 18] have been introduced into S. cer- evisiae strains, as well combining both pathways into the yeast, simultaneously [14]. Recently, the industrial strain CelluX^TM1 [19], a xylose engineered strain with a XI pathway, has been developed.

Interestingly, an unanticipated phenotype that has emerged from xylose strain development is hyper-sensi- tivity of the introduced heterologous metabolic pathways to stressful conditions [2]. Due to the inter-connectivity between metabolism and stress response, strain devel- opment for lignocellulose bioconversion technologies have to simultaneously address both xylose utilization and microbial stresses [10]. Xylose engineered industrial strains are thus the ideal genetic background in which to study the impact of microbial stresses, as well as intro- ducing stress resistance genes. Compared to the XR-XDH route, there is limited knowledge on XI pathway-based xylose utilization for genetically engineered industrial S.

cerevisiae [20]. Furthermore, there is a lack of studies on the interaction of genes involved in inhibitor tolerance and xylose fermentation in these yeasts.

Fermentation strains are subjected to various microbial stresses during lignocellulose bioconversion that include microbial inhibitory compounds generated during phys- icochemical pretreatment of lignocelluloses. The concen- trations of inhibitory compounds fluctuate depending on both biomass composition and pretreatment method used [9, 11, 21]. Furans are degradation products of sug- ars, phenolics are derived from solubilizing lignin and weak acids such as acetic and formic acids are formed during furan degradation and/or de-acetylation of hemi- cellulose [5, 8, 22, 23]. Such microbial inhibitors nega- tively impact growth, fermentation and xylose utilization ability of yeast, which results in sub-optimal ethanol productivity and yields [9, 10, 13, 16]. Given the toxicity

of these compounds, studies have been undertaken to develop yeast strains capable of not only withstanding the harsh conditions associated with lignocellulosic bio- mass fermentations, but to also generate ethanol yields expected in industrial processes [5, 22]. Thus, microbial inhibitor toxicity represents a bottle-neck in lignocel- lulosic bioethanol production and negating these det- rimental inhibitory effects remains to be a fundamental challenge [22, 24, 25].

Inhibitor resistance is characterized as a complex func- tion of multiple genes [10, 26–29], however, relatively few over-expression studies have undertaken to genetically engineer yeast towards multiple inhibitor tolerance phe- notypes. Typical rational design strategies are often lim- ited to only a few genes involved in highly specific in situ detoxification mechanisms, resulting in strains with inhibitor-specific detoxification phenotypes [30–34].

Examples include overexpression of either TAL1, FDH1, or HAA1 for weak acids resistance [30, 35] and ADH6, ADH7 or ARI1 [36–38] for furfural detoxification mecha- nisms, as well as PAD1 and/or FDC1 that are linked to phenolic detoxification [33, 34, 39]. These studies have several shortcomings in that (i) the majority of over- expression studies are conducted in laboratory strains limiting applicability to industrial strains; (ii) strategies are limited by the available knowledge of the molecular genetic basis of resistance phenotypes gained from lab- oratory strains; (iii) studies make use of synthetic cock- tails to simulate industrial stresses, and (iv) very few host strains are proficient xylose utilizers. Thus, more research is required to elucidate gene interaction or synergism within the dynamics of inhibitor resistance phenotypes in industrial strains towards industrial hydrolysates. A cumulative strain development strategy that combines multiple positive gene interactions from various stress response pathways could significantly enhance the yeast stress response towards multi-inhibitor resistance phe- notypes. As a result, this study aims to combine inhibi- tor resistance pathways within a metabolic engineering approach towards the development of efficient multi- inhibitor resistant xylose-utilizing strains.

This study explored the simultaneous overexpression of multiple native gene targets that confer resistance to weak acids, furan aldehydes and phenolic compounds individually, and thereby identify gene combinations that could generate cumulative multi-inhibitor resist- ance phenotypes. ARI1 (NADPH-dependent alde- hyde reductase), PAD1 (flavin prenyltransferase) and TAL1 (transaldolase) were selected since these genes have been implicated in broader multi-inhibitor resist- ance phenotypes [32, 39, 40], whereas FPS1 (aquaglyc- eroporin), FDH1 (formate dehydrogenase 1), ADH6 (NADP-dependent alcohol dehydrogenase 6) and ICT1

(1-acylglycerol-3-phosphate O-acyltransferase) address specific effects attributed to weak acid [30, 41, 42], furan aldehyde [37] and organic solvent stresses [43], respec- tively. The partial deletion of FPS1 and inclusion of ICT1 allowed for the novel regulation of “membrane-modu- lating” genes into resistance phenotypes [44]. Dual gene combinations were constructed to have a multi-inhibitor resistance gene coupled with a target-specific gene in an FPS1 deletion background. Not only does this approach allow for the introduction of genes in novel sequential combinations, but also the assessment of these genes in industrial strains exposed to industrial-like lignocellu- lose fermentations. Ultimately, this improves the current understanding of in situ detoxification of lignocellulose- derived inhibitors for the development of robust xylose- capable S. cerevisiae industrial strains.

Results

Chemical composition of lignocellulose hydrolysate and SSL

Table 1 lists some of the components determined in the liquid fraction of steam-exploded sugarcane bagasse (SCB), as well as SSL from MgO acid sulphite pulping process of mixed hardwood feedstocks. The pretreat- ment as well as sulphite pulping process resulted in the solubilization of the hemicellulosic fraction, but lignin was solubilized more intensely in the case of the sulphite

process. Both liquors were composed of fermentable sug- ars and inhibitors such as weak acids (15.7 and 6.5 g L⁻¹ for SSL and SCB hydrolysate, respectively), furans (2.3 and 2  g L⁻¹ for SSL and SCB hydrolysate, respectively) and phenolics (2  g L⁻¹ and 0.8  g L⁻¹ for SSL and SCB hydrolysate, respectively). In terms of carbohydrates, the main sugar present was xylose with values of about 93 g L⁻¹ and 8.7 g L⁻¹ for the SSL and SCB hydrolysate, respectively. Literature reports xylose as the main sugar present in SSL liquors generated from hardwood feed- stocks (HSSL) [11]. Also, SSL contained considerably more xylose than SCB hydrolysate, as well as significant concentrations of inhibitors, especially acetic acid (15.1 g L⁻¹, about 3 times more than SCB hydrolysate) and phe- nolic compounds (2 g L⁻¹, two times more than in steam explosion hydrolysate). Moreover, the SSL also has other compounds that can act as inhibitors such as SO₂ (used extensively in the wine industry to limit microbial con- tamination at lower pH values) and MgO (inhibits the growth of S. cerevisiae), making hardwood SSL a particu- larly challenging lignocellulosic feedstock [11].

Strain development in lignocellulose hydrolysate and inhibitor tolerance assays

The multi-inhibitor-resistant strain construction strat- egy centred on three rounds of sequential delta integra- tion of double-gene expression cassettes to construct strains overexpressing selected inhibitor resistance genes in different combinations. Overexpression of respective double-gene cassettes was facilitated by homologous recombination of delta-integration cassettes into native delta sequences distributed in the parental CelluX^TM1 yeast genome. Transformants were screened for growth and ethanol yield in lignocellulosic hydrolysate after each round of integration to select the best strains for the next round of transformation, with final transformants assayed for inhibitor tolerance phenotypes.

Before the integration of inhibitor tolerance genes commenced, a partial FPS1 deletion variant of S. cerevi- siae CelluX^TM1 was generated. Eight partial FPS1 dele- tion transformants were selected by screening for higher ethanol yield (g ethanol g⁻¹ total sugar) on 2% YPD sup- plemented with 65% v/v sugarcane hydrolysate (Table 1).

The best performing transformant, CelluX1∆FPS1-C5, yielded 0.41 g  g⁻¹ at 169 h that showed a 5% ethanol yield increment to the parental CelluX^TM1 strain at 0.39 g  g⁻¹ (see Additional file 1: Tables S2, S3). Interestingly, strain CelluX1∆FPS1-C5 also exhibited an increase of 19.8% in formic acid detoxification and higher xylose consump- tion at 53.6% compared to 49.8% of the parental strain.

This partial FPS1 deletion CelluX1∆FPS1-C5 strain was used as the host for the first round of delta integrations.

Table 1 Chemical composition of concentrated hardwood‑SSL and sugarcane hydrolysate

a Hydroxymethylfurfural

b Dihydroxybenzoic acid

Component Hardwood SSL Hydrolysate–

pretreatment liquor

Units

Glucose 14.7 ± 0.02 1.04 ± 0.02 g L⁻¹

Xylose 92.7 ± 1.03 8.70 ± 0.06 g L⁻¹

Furfural 2.08 ± 0.00 1.66 ± 0.01 g L⁻¹

HMF^a 0.21 ± 0.01 0.35 ± 0.00 g L⁻¹

Acetic acid 15.1 ± 0.48 5.76 ± 0.02 g L⁻¹

Formic acid 0.56 ± 0.01 0.74 ± 0.02 g L⁻¹

Cinnamic acid 72.0 ± 18.0 – mg L⁻¹

Ferulic acid 275 ± 21.4 33.8 ± 4.80 mg L⁻¹

3,4‑DHBA^b 46.1 ± 9.25 516 ± 4.22 mg L⁻¹

3–5 DHBA^b 1.05 ± 0.36 0.002 ± 0.39 g L⁻¹

Vanillic acid 116 ± 32.1 16.1 ± 0.56 mg L⁻¹

Syringic acid 308 ± 34.4 29.3 ± 0.23 mg L⁻¹

Vanillin 76.1 ± 9.39 135 ± 3.05 mg L⁻¹

Syringaldehyde 138 ± 14.1 23.1 ± 1.15 mg L⁻¹

Coniferaldehyde 15.9 ± 6.20 16.6 ± 0.77 mg L⁻¹

MgO 17.2 – g L⁻¹

SO₂ 0.6 – g L⁻¹

The results from the best transformant per gene com- bination from the first round of delta integration are listed in Table 2. The first round of transformants often exhibited a trade-off between growth and ethanol yield (see Additional file 2: Figure S1). The pBKD-AA integra- tion cassette with the ARI1 and ADH6 genes that confer furan resistance, in combination with partial FPS1 dele- tion allowed for furan and weak acid resistance. Final ethanol yields of pBKD-AA transformants showed over- all improvement that ranged from 0.34 to 0.38 g  g⁻¹ as compared to the parental CelluX^TM1 strain at 0.33 g  g⁻¹, with a maximum increase in the ethanol yield of 15.8%

(Table 2). Partial deletion of the FPS1 gene in combina- tion with pBKD-AA integrations proved beneficial to inhibitor resistance phenotype in terms of cell growth, as 80% of transformants displayed similar or increased growth (measured in absorbance) compared to the paren- tal strain. The pBKD-AF integration cassette overexpress- ing the ARI1 and FDH1 genes also conferred furan and weak acid resistance. This configuration, however, signifi- cantly decreased the ethanol yield by 4.29–17.2%, com- pared to the parental strain, although an improvement in growth was seen with AF10, showing an increment of 7.92% over the parental strain. The pBKD-AI integra- tion cassette overexpressing the ARI1 and ICT1 genes conferred furan, organic solvent and weak acid resistance however, 62.5% of transformants with this configuration exhibited decreased growth of 12–15%. Likewise, etha- nol yields decreased by 1–7% showing improvement for only one strain (AI1) with an increment of 1.25% over the parental reference, respectively (see Additional file 2: Fig- ure S1).

The pBZD-PA integration cassette overexpressing the PAD1 and ADH6 genes and this configuration conferred resistance to furans, phenolics and weak acids. After the first round of delta integrations, the ethanol yields of the transformants were lower by 2–24.6% or equal to that of the parental strain with six transformants showing

improved growth by 1–18.8% growth increment (see Additional file 2: Figure S1). The pBZD-PF cassette inte- gration overexpresses the PAD1 and FDH1 genes that confer resistance to phenolics and weak acids. These transformants also exhibited no increases in ethanol yields, but showed an increase in growth by 5–31.7% for most of the transformants. A similar trend was observed in the transformants with the pBZD-PI integration cas- sette overexpressing the PAD1 and ICT1 genes, which confers resistance to weak acids, phenolics and organic solvents. Ethanol yields were either lower or similar to the parental strain, whereas growth was either similar or higher than that of the reference strain with PI3 transfor- mant showing a maximum growth increment of 20.9%.

The pBHD-TA integration cassette overexpresses the TAL1 and ADH6 genes that confer furan and weak acid resistance. After the first round of delta integra- tions, the pBHD-TA transformants showed no real dif- ferences in ethanol yields relative to the parental strain with only TA6 showing improvement at 3.6% ethanol yield increment and final growth increment of 18.6%.

Transformants with the pBHD-TF integration cassette overexpressing the TAL1 and FDH1 genes conferring weak acid resistance exhibited similar or higher etha- nol yields relative to parental strain with TF2 showing the highest increment in yield at 16.9%. The pBHD-TI integration cassette overexpressing the TAL1 and ICT1 genes that confers resistance to weak acids and organic solvents proved to be detrimental to both the ethanol yield and growth as transformants exhibited significant decreases in both ethanol yield (8.4–35.2%) and growth (20.4–38.2%).

After the first stage of delta integration, the pBHD-TF and pBKD-AA integration cassettes generated transfor- mants with more than 10% increment on ethanol yield relative to the parental strain. Therefore, these com- binations were selected to continue into the second round that involved the integration of the pBHD-TF and

Table 2 Performance of the best transformant per gene combination after 1st stage of strain development

Gene combination Strain Resistance phenotype EtOH yield %

increment Growth (OD₆₀₀)

% increment

ARI + ADH6 AA6 Furans + weak acids 15.80 19.50

ARI1 + FDH1 AF10 Furans + weak acids − 5.55 7.92

ARI1 + ICT1 AI1 Furans + weak acids + organic solvents 1.25 − 12.00

PAD1 + ADH6 PA7 Phenolics + furans + weak acids − 1.97 3.40

PAD1 + FDH1 PF5 Phenolics + weak acids − 15.40 31.80

PAD1 + ICT1 PI3 Phenolics + weak acids + organic solvents − 3.58 20.90

TAL1 + ADH6 TA6 Furans + weak acids 3.60 18.60

TAL1 + FDH1 TF2 Weak acids 16.90 6.20

TAL1 + ICT1 TI10 Weak acids + organic solvents − 8.34 − 22.30

pBKD-AA integration cassettes into S. cerevisiae AA6 (resulting in ATF transformants) and TF2 (TFA transfor- mants), respectively. The second round of transformants also resulted in a trade-off between growth and ethanol yield (see Additional file 2: Figure S2). Compared to the parental strain, 70.8% of the ATF transformants favoured growth with ATF13 showing an improved growth incre- ment at 17.3%. Conversely, 58.3% of the TFA transfor- mants exhibited increased ethanol yields. Interestingly, when the concentration of hydrolysate in YPDX reduced from 65% v/v to 50% v/v, no significant differences were observed in resistance phenotypes between parental strain and transformants, indicating possible phenotypic plasticity in transformants.

In the third and final round of delta integration, plas- mid pBZD-PI was integrated into the ATF13 (AP trans- formants) and TFA7 (TP transformants) strains, thus strains overexpress six genes in the FPS1 partial deletion background. Subsequent transformants exhibit resist- ance to weak acids, furans and phenolic compounds. The strains were evaluated on growth (absorbance), fermen- tation ability (ethanol yield) and inhibitor detoxification (% conversion). S. cerevisiae CelluX^TM1 was used as an industrial and parental reference strain, whereas ATF13

and TFA7 were used as additional parental reference strains. All transformants showed a significant improve- ment in growth compared to the parental CelluX^TM1 strain, during fermentations with 2% SC-X media sup- plemented with 65% v/v sugarcane hydrolysate at pH 5 and spiked with 20 g L⁻¹ of furfural and 20 g L⁻¹ of for- mic acid (Fig. 1A). The growth profiles of the AP and TP transformants surpassed the performance of the indus- trial CelluX^TM1 strain. Interestingly, TFA7 with only two gene cassettes did as well as the final transformants. AP1 and AP4 showed the highest growth (OD600), whereas TP1 was the best performer from the TP transformants (Fig. 1A). Only AP1 and TP1 showed a significant differ- ence in growth between 120 h versus 168 h. As expected, ethanol concentrations were very low, ranging from 1.6 to 2.4  g L⁻¹ given the extreme toxicity of fermentation media (data not shown).

The transformants showed a noteworthy differ- ence in inhibitor detoxification phenotypes for formic acid (present as formate in the medium at pH 5) and furfural (Fig. 1B). Although AP4 showed the highest OD600 at 168 h, the growth could not be linked back to an improved inhibitor resistance phenotype. In con- trast, the TP1 strain showed the highest detoxification

0.0 0.5 1.0 1.5 2.0 2.5

Biomass (OD600)

120 h 168 h

A

C

B

AA6 TF2 PI3 TFA7 AP1 TP1

2% SC Control Furfural Weak acids Syringaldehyde Temperature 37°C CelluX^TM1

Strains Inhibitor

0 5 10 15 20 25

Formic acid Furfural

% Conversion

CelluX^TM1 TFA7 AP1 AP4 TP1

Strains

Fig. 1 The performance of final transformants in A 2% SC‑X supplemented with 65% v/v sugarcane hydrolysate spiked with 20 g L⁻¹ furfural and formic acid, with a straight line indicating benchmark performance of CelluX^TM1 parent and B the % conversion of inhibitors. C A spot chart illustrates transformant performance to various inhibitors/stressors at OD₆₀₀ (from left to right) 5,1,0.5, 0.1 and 0.01 in 2% SC only or supplemented with 1 g L⁻¹ furfural, 6 g L⁻¹ acetic acid and 0.8 g L⁻¹ formic acid, or 0.6 g L⁻¹ syringaldehyde, and growth at 37 °C

phenotype with an average of 13.4% and 19% conver- sion of formic acid (formate) and furfural, respectively, at 168 h (Fig. 1B). CelluX^TM1 as parental and industrial reference strain, however, showed poor growth and poor inhibitor detoxification with no formic acid con- verted and only 3% of the furfural detoxified. TFA7 and AP1 transformants also surpassed the parental and industrial reference strain with improved detoxification phenotypes.

The copy number of the gene inserts were determined via qPCR analysis. The TFA7, TP1 and AP1 transfor- mants each have 1 insert of the TF (TAL1 + FDH1) and AA (ARI1 + ADH6) double-gene constructs, i.e. 1 addi- tional copy of the native TAL1, FDH1, ARI1 and ADH6 genes. Both TP1 and AP1 transformants also have 6 addi- tional copies of native PAD1 and ICT1 genes (see Addi- tional file 1: Table S4). The copies numbers of the inserts between the transformants were similar. However, the observed phenotypes of the TFA7, TP1 and AP1 transfor- mants were different as the AP1 strain performed poorly compared to the more robust TFA7 and TP1 strains.

Two different assay methods were applied to evaluate inhibitor tolerance phenotypes, i.e. inhibitor tolerance plate assays (pH 4.0–4.5, no pH control) and enzymatic assays. Plate assays showed variations between inhibitor phenotypes within the three stages of strain development (Fig. 1C). In particular, the PI3 strain shows increased susceptibility to weak acid stress (6 g L⁻¹ acetic acid and 0.8  g L⁻¹ formic acid) when no pH control was imple- mented (pH < 5) and this phenotype was confirmed in AP1 and TP1 transformants with pBZD-PI inserts in the third round of delta integrations (Fig. 1C), although this integration improved resistance to the phenolic syringal- dehyde. AP1 was the only transformant showing resist- ance to 1 g L⁻¹ furfural when critical mass was present [45]. The in  vitro activities of detoxification enzymes were assayed to determine inhibitor detoxification poten- tial of transformants. Detoxification was measured as the decrease in substrate, i.e. furfural, cinnamic acid or for- mic acid due to enzymatic degradation (see Additional file 1: Table S5). No significant differences were observed between transformants and parental reference in furfural assays. In the cinnamic acid assays, PI3 and TP1 transfor- mants exhibited enhanced in vivo cinnamic acid detoxifi- cation activity. Similarly, formic assays showed AP1 and TP1 transformants to have enhanced formic acid detoxi- fication phenotypes, relative to control.

Detoxification phenotypes in simulated/synthetic inhibitor cocktail fermentations

The TFA7, AP1 and TP1 transformants were subjected to fermentations in 2% SC-X media supplemented with either, 5  g L⁻¹ furfural plus 0.5  g L⁻¹ HMF, 6  g

L⁻¹ acetic—plus 0.81 g L⁻¹ formic acid, or 1 g L⁻¹ cin- namic acid to ascertain detoxification phenotypes of the gene combinations to specific microbial inhibitors groups. The S. cerevisiae CelluX^TM1 strain was used as an industrial and parental reference. The transformants from the first round of integration (AA6, TF2 and PI3) were used as secondary controls to determine if second and third-round delta-integrated transformants also exhibit phenotype from first integrations, i.e. cumula- tive phenotypes.

In fermentations with 1 g L⁻¹ cinnamic acid, there were no differences observed between parental CelluX^TM1 strain and TFA7 or TP1 transformants. However, the PI3 transformant exhibited an enhanced cinnamic acid detoxification phenotypes (Fig. 2A). The AA6, TFA7, AP1 and TP1 transformants showed marked improvement in furfural detoxification phenotype when compared to parental and industrial reference strains. All transfor- mants exhibited a decrease in the lag phase of 24 h com- pared to the 48  h for the parental strain, with furfural detoxified within the said time period (Fig. 2B). At 48 h, transformants exhibited ethanol yields ranging from 0.25 to 0.29 g  g⁻¹, whereas the parental control fermentation yield was below 0.1 g  g⁻¹ (Fig. 2D). Glucose was depleted within 48 h versus 72 h for the parental CelluX^TM1 refer- ence strain with no significant differences in ethanol yield for TFA1 and AA6 strains versus the reference strain at 72 h. No significant differences in CelluX^TM1 and TFA7 fermentation performances were observed with weak acid exposure, however, this fermentation confirmed AP1 and TP1 strains are more susceptible to weak acids due to the pBZD-PI insert (Fig. 2C).

Fermentations with inhibitor cocktail were conducted with 2% SC supplemented with an inhibitor cocktail (IC) based on the composition of SSL (Table 1). Blank media supplemented with the cocktail was used as control to account for the evaporation of volatiles. Two differ- ent carbon sources were used to determine the possible effect the carbon source may have on resistance pheno- types, given the sensitivity of the introduced heterolo- gous pathways to fermentation stresses. As anticipated, strains showed a significant difference in observed inhibitor resistance phenotypes in fermentations with glucose and xylose versus xylose only as carbon source (Fig.  3). Transformants in xylose-only fermentations showed poor detoxification phenotypes with < 5% of inhibitor compounds detoxified. In fermentations with both glucose and xylose, differences in detoxification phenotypes between transformants and the CelluX^TM1 reference strain were observed. The AP1 transformant outperformed strain CelluX^TM1 for furfural detoxifi- cation, whereas the TP1 transformant outperformed both AP1 and CelluX^TM1 for formic acid detoxification.

Overall, transformants exhibit enhanced detoxification phenotypes compared to CelluX^TM1 parent.

Hardwood SSL fermentations and final growth rate of transformants

Various concentrations of untreated SSL were used to characterize strain performances in lignocellulose fer- mentations in terms of consumption of glucose and xylose, ethanol concentration, ethanol yield and ethanol productivity (Table 3). In 2% SC media supplemented with 40% v/v concentrated SSL at pH 5.0, glucose was depleted within 72  h, however, xylose consumption was less than 10% for all strains with the AP1 transfor- mant showing no xylose consumption. Ethanol yields at 72 h showed that the TP1 transformant with a yield of 0.255 g  g⁻¹ performed better than the CelluX^TM1 strain with yield a of 0.228 g  g⁻¹ (Table 3), an 11.8% increment in yield above parental control. In 2% SC supplemented with 80% v/v concentrated SSL at pH 4.5, both parental and transformant strains showed no growth, however, strains appeared to be metabolically active as seen by the consumption of sugars (Table 3). Glucose consump- tion for all strains exceeded 10%, with CelluX^TM1, TFA7 and TP1 consuming 15.1%, 15.8% and 16.3%, respec- tively. However, ethanol concentrations remained

below 1  g L⁻¹, with only CelluX^TM1, TFA7 and TP1 strains producing ethanol, at 0.51, 0.66 and 0.48 g L⁻¹, respectively.

The growth kinetics of the final transformants were characterized in 2% YPDX and YPX at pH 5.0 in terms of consumption of glucose and xylose, ethanol con- centration, ethanol yield, ethanol productivity, meta- bolic ethanol yield and the maximum specific growth rate (µmax) (Table 3). In 2% YPDX, the TFA7 and TP1 transformants exhibited both increased ethanol yield and productivity at 0.428 g  g⁻¹ and 0.713 g L⁻¹  h⁻¹ and 0.432  g  g⁻¹ and 0.720  g L⁻¹  h⁻¹, respectively. The co- fermentation of glucose and xylose was only reduced for strain API1 during YPDX fermentations, unlike the low xylose utilization for all strains seen in SSL (0 – 6% xylose consumption), highlighting the pronounced effect of inhibitors on % xylose consumption. Further- more, xylose consumption in 2% YPX decreased from 100% of the reference CelluX^TM1 strain to 42.3 – 90.1%

for transformants, indicating that the strain modifica- tion impacted negatively on xylose consumption. This was confirmed in hardwood-SSL fermentations with xylose as the main carbon source, where transformants exhibited lower % xylose consumption versus reference CelluX^TM1 strain.

0 2 4 6 8 10 12 14 16

0 24 48 72 96 120

Biomass (OD600)

Hours

CelluX^TM1 AA6 TFA7 AP1 TP1

0.0 0.1 0.2 0.3 0.4 0.5

0 5 10 15 20

CelluX^TM1 TFA7 AP1 TPI1

g(EtoH)g-1 (Total Sugar)

Strains

OD₆₀₀ Acetate g L^-1 Formate g L-1 EtOH yield 0

10 20 30 40 50 60 70 80

CelluX^TM1 PI3 TFA7 TP1

%Cinnamicacid conversion

Strains

0.00 0.05 0.15 0.20 0.25 0.30 0.35 0.40 0.45

CelluX^TM1 AA6 TFA7 AP1 TP1 Strains

48h 72h

A B

C D

OD600and Compounds(g L-1) g(EtoH)g-1 (Total Sugar)

Fig. 2 Performances of final transformants in 2% SC‑X fermentations supplemented with either, A 1 g L⁻¹ cinnamic acid, B 5 g L⁻¹ furfural or C 6 g L⁻¹ acetic and 1 g L⁻¹ formic acid. D The ethanol yields of various strains in 2% SC‑X fermentations supplemented with 5 g L⁻¹ furfural at 48 versus 72 h

Discussion

Efficient carbohydrate bioconversion of lignocellulose hydrolysates remains a challenge given the numerous microbial stresses and inhibitory compounds produced during pre-treatment, despite the progress on the devel- opment of industrial yeast strains with the ability of co- fermenting glucose and xylose. Overexpression of genes involved in detoxification of inhibitors could be beneficial for improving yeast tolerance. This study used a rational engineering strategy based on sequential delta homolo- gous integrations of dual expression cassettes, allowing the introduction of several copies into the genome of the yeast. Novel target gene combinations were selected to not only modulate the cell plasma membrane against inhibitor influx, but also to direct intracellular detoxifica- tion of inhibitors and strengthen carbon/xylose metabo- lism. One copy of the FPS1 gene in a xylose-capable (XI) industrial strain S. cerevisiae CelluX^TM1 was deleted, whereas the TAL1, PAD1, FDH1, ICT1, ARI1 and ADH6 genes were overexpressed to develop a range of multi- inhibitor-resistant strains. The impact of gene combi- nations on the development of cumulative inhibitor resistance phenotypes was evaluated.

The first step of our study was the disruption of the FPS1 aquaglyceroporin gene, which encodes a channel

protein responsible for glycerol efflux and intake of ace- tic acid. The deletion of FPS1 in industrial S. cerevisiae strains has proven to substantially increase both growth and ethanol yield (10–45% improvement) on glucose media under acetic acid stresses [41] and improve xylose fermentation as seen by 3–10% improvement on ethanol yield [42]. The partial deletion of the FPS1 gene in our study, however, resulted in a moderate increase on the ethanol yield (~ 5% increment), that could be ascribed to the partial deletion and/or differences in the genetic background of the strain and media (carbon source: glu- cose and xylose; concentration of acetic acid 2.9 g/L, pH 5).A CelluX1∆FPS1-C5 transformant was used as recipi- ent strain for the sequential delta integration of nine different gene combinations. These combinations were assessed for ethanol yield and biomass formation (growth measured at OD₆₀₀) during fermentations supplemented with 65% v/v sugarcane bagasse hydrolysate (Table 1).

Compared to parental strains, there was generally a trade-off between the growth and the ethanol yield of the transformants from the first round of delta integrations (see Additional file 2: Figure S1). Nevertheless, some combinations of genes resulted in significant improve- ment on ethanol yield (ARI1 + ADH6 and TAL1 + FDH1)

0 5 10 15 20

% Conversion Glucose + Xylose

Furfural Formic acid

Strains 0

1 2 3 4 5 6

CelluX^TM1 TFA7 AP1 TP1

B

% Conversion Xylose only

Fig. 3 The % conversion of 5 g L⁻¹ furfural and 0.8 g L⁻¹ formic acid by transformants in 2% SC fermentations supplemented with synthetic IC with A glucose and xylose as carbon source or B xylose as only carbon source at 120 h

while other combinations were detrimental for both parameters (TAL1 + ICT1) (Table 1). The benefits of ARI1 and ADH6 overexpression on biomass and cell viability have been documented [37, 46]. Similarly, our results are in line with those of Sanda et al. [47] where the overexpression of both TAL1 and FDH1 resulted in improved ethanol production in xylose-fermenting labo- ratory strains.

None of the transformants screened could ferment the sugars when the SCB hydrolysate concentration in the media was increased from 65 to 75% v/v (data not shown), that suggests selecting 2–3 genes to improve inhibitor resistance fell short of multi-inhibitor resistance phenotypes. Although, it is interesting to note that the inhibitor-specific combinations AA (ARI1 + ADH6) for furans and TF (TAL1 + FDH1) for weak acids were the only combinations to improve both ethanol yields and biomass in 65% v/v hydrolysate fermentations. Inhibitor- specific combinations with different specificities may have a cumulative effect in constructing multi-resistance phenotypes. The AA and TF combinations were used in subsequent strain development, which resulted in strains with these specific gene combinations, but in alternate integration sequences. The PI (PAD1 + ICT1) combina- tion was included only for the final round of integrations

as it did not improve ethanol yields but did improve growth (Table 2).

The second round of delta integrations also resulted in a trade-off between ethanol yield and biomass growth, with the ATF transformants more prone towards bio- mass (2–17% increment in 70.8% of the transformants) while the TFA transformants were more inclined to ethanol yield improvement (5–24% increment in 58.3%

of the transformants) (see Additional file 2: Figure S2).

It was also found that transformants presented pheno- typic plasticity, with no differences in growth or ethanol yield compared to parental strain when the toxicity of the media was reduced. The selected transformant for the next round of integration, TFA7, was also able to grow in the presence of syringaldehyde at low cells concentration (Fig. 2C).

The third round of delta integration resulted in trans- formants with different combinations of the selected six genes in a partial FPS1 deletion background. Of particu- lar interest, was whether subsequent additions of inhibi- tor-specific gene combinations had a cumulative effect or build-up towards multi-inhibitor resistance phenotypes.

The performance of the final transformants was evalu- ated during fermentations supplemented with inhibitors (single inhibitor or in a cocktail) and different source of Table 3 Fermentation kinetic parameters of recombinant S. cerevisiae strains and control strains

a Not detected/determined

Strain Glucose

cons. % Xylose

cons. % Ethanol Metabolic

yield % Media pH µMax (h⁻¹) g L⁻¹ Y P/S g g⁻¹ g L⁻¹  h⁻¹

YPDX t=24 h

CelluX^TM1 100 100 17.0 ± 0.13 0.424 ± 0.00 0.707 ± 0.01 82.8 5 0.572

TFA7 100 100 17.1 ± 0.13 0.428 ± 0.00 0.713 ± 0.01 83.5 5 0.557

AP1 100 45.7 12.2 ± 0.09 0.305 ± 0.00 0.508 ± 0.00 82.1 5 0.513

TP1 100 100 17.3 ± 0.08 0.432 ± 0.00 0.720 ± 0.00 84.4 5 0.545

YPX t=24 h

CelluX^TM1 – 100 8.18 ± 0.05 0.396 ± 0.02 0.341 ± 0.00 77.7 5 0.412

TFA7 – 90.1 7.52 ± 0.25 0.364 ± 0.01 0.313 ± 0.01 78.3 5 0.364

AP1 – 42.3 3.81 ± 0.02 0.184 ± 0.00 0.159 ± 0.00 73.7 5 0.442

TP1 – 88.5 7.47 ± 0.35 0.362 ± 0.02 0.311 ± 0.01 78.9 5 0.372

40% v/v SSL _{t=72 h}

CelluX^TM1 100 5.8 11.6 ± 0.93 0.228 ± 0.00 0.161 ± 0.01 – 5 –

TFA7 100 3.0 11.7 ± 0.71 0.225 ± 0.01 0.162 ± 0.01 – 5 –

AP1 100 0 11.8 ± 1.06 0.239 ± 0.00 0.160 ± 0.02 – 5 –

TP1 100 1.28 12.2 ± 1.17 0.255 ± 0.03 0.169 ± 0.02 – 5 –

80% v/v SSL _{t=241 h}

CelluX^TM1 15.1 1.74 0.51 ± 0.03 n.d.^a – – 4.5 –

TFA7 15.8 1.88 0.66 ± 0.03 n.d.^a – – 4.5 –

AP1 10.6 0.76 n.d.^a – – – 4.5 –

TP1 16.3 1.55 0.48 ± 0.05 n.d.^a – – 4.5 –

sugars. The transformants were subjected to fermenta- tions with synthetic inhibitor cocktails, as this allowed for a more controlled assessment to determine if phe- notypes for specific inhibitor resistance could be linked to specific gene combinations. Compared to the paren- tal strain, these transformants showed improved growth during fermentations supplemented with 65% v/v hydro- lysate (Fig. 1A) and proved to be able to detoxify furfural and formic acid despite their high concentration in the media (20  g/L) (Fig. 1B). Final transformants presented strong furan resistance phenotypes with a 24-h reduction of the lag phase in synthetic media containing only fur- fural (Fig. 2B), and improved conversion of furfural when the fermentations were carried out in mixed synthetic inhibitor cocktail with glucose and xylose (AP1, Fig. 3A).

The inclusion of the PAD1-ICT1 combination in both the first and the third round of integration, however, seemed to increase the sensitivity of the yeast towards weak acids at concentrations found in the sugarcane bagasse pretreatment liquor (Table 1), as inferred from the inhibitor tolerance assays (Fig. 1C) and drop in growth in synthetic inhibitor media containing only weak acids (Fig. 2C). This could explain that the TFA7 transformant, that only contains two gene cassettes, was able to outperform CelluX^TM1 and, in some cases, TP1 and AP1 as well. Alternatively, we speculate that this could also be linked to a lower metabolic burden as compared to TP1 and AP1 (Table 3). In terms of resist- ance towards cinnamic acids, there was no significant difference between the parental strain and the last set of transformants (Fig. 2A). In contrast, this resistance phenotype was evident in the first round of integration for the PAD1-ICT1 combo (PI3, Fig. 2A) together with increased susceptibility to higher temperatures (Fig. 1C).

The reduced thermo-tolerance could be due to an excess fluidity of the membrane caused by a higher proportion of unsaturated fatty acids incorporated by the ICT1 pro- tein (1-acylglycerol-3-phosphate O-acyltransferase) [43].

The tolerance of the yeast towards lignocellulosic- derived inhibitors is also dependent on the carbon source in the fermentation media, and xylose metabolism is much more susceptible than glucose’s [2]. However, there is limited information on the possible interac- tion between genes conferring inhibitor tolerance, and the genes associated with the XI pathway, especially in industrial strains of S. cerevisiae. Our results demon- strate the inter-linkage between carbon metabolism and microbial inhibitor resistance. During fermentations with inhibitor cocktail media containing xylose as only carbon source, the conversion of furfural was decreased drasti- cally in all the strains (Fig. 3B). However, this reduction was not as severe in the case of the transformants (2.4–4 times less) compared to the parental strain (7.4 times

less) (Fig. 3B). Invariably, the true test of inhibitor resist- ance is fermentation ability exhibited with lignocellulose hydrolysates. Fermentations in SSL presented a unique challenge to the strain development as it is both xylose rich and contain microbial inhibitors unique to the paper and pulp production process such as lignosulfonates and high concentrations of Ca⁺² or Mg⁺² ions besides the typical compliment of weak acids, furans and phenolics (Table 1).

In 2% YPD supplemented with 40% v/v SSL, transfor- mants TP1 and AP1 outperformed strain CelluX^TM1 (4.8% increment on ethanol yield for API1, 11.84% incre- ment on ethanol yield for TP1), confirming enhanced inhibitor resistance phenotypes (Table 3). Given the poor performances of strains in xylose-only fermentations sup- plemented with inhibitor cocktail, it was expected that strains would not be able to tolerate SSL well. However, this improvement was observed despite a reduction in the xylose consumption compared to the parental strain, especially for the API1 transformant (Table 3). However, the 80% v/v SSL YPD media proved too toxic for all the strains, but it should be noted that the strains were spe- cifically developed with furans, weak acids and phenolic compounds in mind. Hardwood-SSL contains atypical microbial inhibitors such as MgO, lignosulfonates and SO2 which the industrial yeast strains are as yet unable to tolerate at such high concentrations (Table 1).

Unexpectedly, the ethanol production from xylose in the selected transformants was also reduced when fer- mentations were carried out with no inhibitors present (Table 3, values for YPX). Nonetheless, the transformants containing the TAL1 + FDH1 combination from the first integration (TFA7, TP1) were less influenced with about 10% reduction on xylose consumption compared to an almost 58% reduction in the AP1 transformant. The positive synergism of the overexpression of these two genes has been documented for a recombinant xylose- fermenting S. cerevisiae laboratory strain, i.e. the ethanol production from xylose was improved, despite the media containing both acetic acid (1.8  g L⁻¹) and formic acid (0.96 g L⁻¹) [47].

Different gene interactions between gene combinations with no differences in gene copy numbers were observed, suggesting there might also have been a possible “posi- tion effect” influencing the multi-inhibitor resistance phenotypes. It is tempting to speculate that the initial or first integration events exhibited a more dominant phe- notype because the first cassettes integrated into highly active and assessable sites, whereas subsequent inte- grations were relegated to less active areas, a so-called

“positional effect”. As such, the transformation efficiency was reduced with each integration cycle. This positional effect could explain the higher furfural conversion of

API1 (Fig. 3A), and the interaction of TAL1 + FDH1 on xylose metabolism when the combination was the first integration cassette (Table 3: TFA7, TP1), as well as the increment of conversion of cinnamic acids only when the PAD1 gene was on the first integration cassette (Fig. 2A).

Additional study into advanced strain development strategies for the manipulation of complex phenotypes, such as microbial inhibitor resistance with a minimal detrimental impact on xylose fermentation on XI engi- neered industrial yeasts is required. Furthermore, eluci- dating how gene location, gene dosage and copy number influence exhibited phenotypes, would improve the tun- ing of the transformation process by adjusting the DNA concentration and/or sequence of integration accord- ing to genes (function, size) [48]. Combining targeted rational engineering with techniques such as evolution- ary engineering or genome shuffling may pave the way forward in the manipulation of complex phenotypes [49].

Conclusions

The efficient conversion of sugars (glucose and xylose) in the presence of microbial inhibitors for lignocellulose- derived biofuels production still remains a challenge.

This study aimed to evaluate novel gene combinations that confer resistance to multiple inhibitors (cumulative resistance phenotypes) in recombinant xylose-capable industrial yeast strains. The sequential delta-integration of these genes resulted in strains with improved toler- ance towards furans and formic acid, but these acquired abilities somehow negatively influenced the xylose con- sumption capacity of the yeast. Nevertheless, there were combinations of genes where this impact was minimal, specifically when the sequence of integra- tions was TAL1 + FDH1 followed by ARI1 + ADH6 (and PAD1 + ICT1). Despite the reduced xylose fermentation, selected strains could outperform the parental strain when grown on synthetic media supplemented with 40%

(v/v) xylose-rich SSL hydrolysate. This study highlighted inhibitor resistance as a complex phenotype and contrib- utes towards developing advanced strain development techniques based on positive gene interactions/mecha- nisms to develop “hardened” multi-inhibitor resistance xylose-capable S. cerevisiae strains.

Materials and methods

Microbial strains and culture conditions

S. cerevisiae CelluX^TM1 (Leaf by Lesaffre, France) was selected as an industrial strain for rational yeast engi- neering. S. cerevisiae CelluX^TM1 and transformants were routinely cultivated, selected and screened by using YPD (20 g L⁻¹ glucose, 10 g L⁻¹ yeast extract and 20 g L⁻¹ peptone; Merck—Darmstadt, Germany) media sup- plemented with 300–400  µg  mL⁻¹ of the appropriate

antibiotics or combination of antibiotics namely; hygro- mycin B (Calbiochem, San Diego, USA), geneticin (Mel- ford laboratories, Ipswich, UK) and zeocin (Melford, Ipswich, UK). Strains were pre-cultured in synthetic complete (SC-X) media at pH 5 containing 20 g L⁻¹ glu- cose and 20 g L⁻¹ xylose, 5 g L⁻¹ (NH4)2SO4, 1.67 g L⁻¹ YNB w/o amino acids, 3  g L⁻¹ KH2PO4 and 100  mM potassium phthalate, supplemented with 20% inhibitor cocktail (20%-IC) containing 0.2  g L⁻¹ cinnamic acid, 0.1 g L⁻¹ HMF, 1.5 g L⁻¹ furfural, 1.2 g L⁻¹ acetic acid and 0.16  g L⁻¹ formic acid (Sigma Aldrich, St. Louis, USA). Pre-cultures were incubated at 30 °C and shaking at 200  rpm. Growth curves of select strains were con- ducted in YPD and YPDX (20  g L⁻¹ glucose, 10  g L⁻¹ yeast extract, 20 g L⁻¹ peptone, 20 g L⁻¹ xylose; Merck, Darmstadt, Germany) media, incubated at 30  °C, shak- ing at 200  rpm and sampled at 3-h intervals for 24  h.

Escherichia coli DH5α (Life Technologies-CA, USA) was used for plasmid propagation and cloning. E.coli trans- formants were cultivated at 37 °C in Luria–Bertani (LB) media (1% tryptone, 0.5% yeast extract, 1% NaCl; Merck, Darmstadt, Germany) supplemented with 100  µg  mL⁻¹ ampicillin (Roche, Johannesburg, South Africa).

Construction of plasmids

Standard protocols for DNA manipulation were fol- lowed [50]. Genomic DNA was extracted from S. cerevi- siae BY4742∆FPS1 [51] and used as template DNA for amplification of target genes open reading frame. Target genes ARI1, ADH6, FDH1, ICT1, PAD1, and TAL1 were amplified via PCR using the Phusion^® high-fidelity DNA polymerase (New England Biolabs, Ipswich, USA) and appropriate primers (Table 4) on an Applied Biosystems 2720 thermocycler (Life Technologies, CA, USA) accord- ing to the manufacturers’ recommendations. The primers introduced PacI and AscI restriction sites required for directional cloning into the delta-integration plasmids, pBZD [52], pBKD [52] and pBHD [53]. PCR products were initially ligated into the pCLoneJET 1.2 commercial vector (Thermo Scientific, Waltham, USA) according to the manufacturers’ guidelines. Gene sequences were veri- fied using the dideoxy chain termination method and an ABI PRISM^™ 3100 genetic analyser (Applied Biosystems, Waltham, USA) at Central Analytical Facility (CAF) of Stellenbosch University.

The first gene expression cassettes were constructed by directional cloning of ARI1, TAL1 or PAD1 into plasmids pBKD1, pBHD1 or pBZD1 (D1), respectively, contain- ing the constitutive PGK1 gene promoter and termina- tor sequences. Secondary gene expression cassettes were constructed by directional cloning of ADH6, FDH1 or ICT1 into plasmid pBKD2 (D2) containing the consti- tutive ENO1 gene promoter and terminator sequences.

Double gene expression cassettes were generated by sub- cloning the pBKD2 ENO1_pt gene cassettes as a SpeI/NotI fragment into corresponding pB(K/H/Z)D1 plasmids to yield single delta plasmids with both PGK_pt and ENO_pt expression cassettes (see Additional file 1: Table S1). All plasmids used and constructed in the study are listed in Table 5.

Yeast transformation and screening FPS deletion stains

The first step of the rational engineering strategy was to disrupt FPS1 to generate FPS1 deletion strains using plasmid pYFCUP1 [41]. Plasmid DNA was propagated and extracted using cetyltrimethylammonium bromide (CTAB) plasmid extraction protocol [50], and used as a template to amplify the FPS1L-CUP1-FPS1R insert by PCR using Phusion^® high-fidelity DNA polymerase (New England Biolabs, Ipswich, USA) and appropriate prim- ers. The 3030 bp linear PCR product was separated on 1%

agarose gel to confirm the insert. The PCR product was then purified using GeneJet PCR purification kit (Thermo Scientific, Waltham, USA) and transformed into S. cer- evisiae CelluX^TM1 by electroporation using a Bio-Rad Gene-Pulser Apparatus (1.4 kV, 200 OHMS, and 25 µF).

Transformants were incubated in 2% YPDS (20 g L⁻¹ glu- cose, 10 g L⁻¹ yeast extract, 20 g L⁻¹ peptone, 1 M sorbi- tol) media at 30 °C for 4–5 h, plated on YPDS agar plates supplemented with 7  mM and 8  mM CuSO₄ and incu- bated for 72  h at 30  °C. Successful transformants were confirmed with PCR using FPS1-L forward and CUP1-L

reverse primers. Partial FPS1 deletion was confirmed and attributed to the aneuploidy nature of parental strain.

The partial FPS1 deletion strains were screened in 70-mL fermentations using 2% YPDX supplemented with sugar- cane pre-treatment liquor/hydrolysate to a concentration of 50% v/v. Fermentations were sampled at 24-h intervals for 7 days and samples were analysed for fermentations products as described in HPLC analysis sector.

Strain construction

The first round of delta integration involved the trans- formation of nine distinct double-gene expression cas- settes into the partial FPS1 deletion CelluX^TM1 strain.

All integration plasmids were digested with either Bst11071 or XhoI (Thermo Scientific—Waltham, USA) according to the manufacturer recommendations and transformed into partial FPS1 deletion CelluX^TM1 strain by electroporation (1.4  kV, 200 OHMS, and 25 µF) using a Bio-Rad Gene-Pulser Apparatus. Trans- formants were recovered on 2% YPD supplemented with appropriate antibiotics and confirmed via PCR using PGKseq-L (D1) and BKDENOpt-L (D2) as for- ward primers in conjunction with gene-specific reverse primers to confirm the complete double-gene insert (Table 4). The transformation frequency of the delta plasmids allowed for a range of copies to be integrated, thus preliminary plate screenings were conducted to identify transformants with higher copy numbers. The plate assays were based on antibiotic resistance, i.e.

higher copy numbers transformants exhibit increased Table 4 Primers used in the study

Restriction sites underlined

Gene Primer Primer 5–3′ sequence References

ADH6 ADH6‑F GCGCC TTA ATT AAATG TCT TAT CCT GAGAA This study

ADH6 ADH6‑R GTTA GGC GCG CCCTA GTC TGA AAA TTC This study

PAD1 PAD1‑F GGCC TTA ATT AAATG CTC CTA TTT CCA AGA AG This study

PAD1 PAD1‑R GATT GGC GCG CCTTA CTT GCT TTT TATT This study

ICT1 ICT1‑F GGCC TTA ATT AAATG TGG ACA AAC ACT TTC AAA TGG This study

ICT1 ICT1‑R GATT GGC GCG CCTTA CTT TGA CAG GAAC This study

ARI1 ARI1‑F GGCC TTA ATT AAATG ACT ACT GAT ACC ACT G This study

ARI1 ARI1‑R GATT GGC GCG CCTTA GGC TTC ATT T This study

FDH1 FDH1‑F GGCC TTA ATT AAATG TCG AAG GGA AAGG This study

FDH1 FDH1‑R GATC GGC GCG CCTTA TTT CTT CTG T This study

TAL1 TAL1‑F GCGC TTA ATT AAATG TCT GAA CCA GCTC This study

TAL1 TAL1‑R GATA GGC GCG CCTTA AGC GGT AAC TTTC This study

FPS1 FPS1‑L CCG AAG CTT ATG AGT AAT CCT CAA AAA GC [41]

FPS1 FPS1‑R CCA GAG CTC TCA TGT TAC CTT CTT AGC ATT [41]

CUP1 CUP1‑L CTT GGT ACC TGG GCG CTA TAC GTG CAT ATG [41]

PGK1 PGKseq‑L CTA ATT CGT AGT TTT TCA AGT TCT TAG ATG C [54]

ENO2 BKDENOpt‑L TCA GTC TAG AGC GGC CGC CTT CTA GGC GGG TTATC This study